Network Loss Research Articles

A simplified predictive framework for cost evaluation to fault assessment using machine learning

<span lang="EN-US">Software engineering is an integral part of any software development scheme which frequently encounters bugs, errors, and faults. Predictive evaluation of software fault contributes towards mitigating this challenge to a large extent; however, there is no benchmarked framework being reported in this case yet. Therefore, this paper introduces a computational framework of the cost evaluation method to facilitate a better form of predictive assessment of software faults. Based on lines of code, the proposed scheme deploys adopts a machine-learning approach to address the perform predictive analysis of faults. The proposed scheme presents an analytical framework of the correlation-based cost model integrated with multiple standards machine learning (ML) models, e.g., linear regression, support vector regression, and artificial neural networks (ANN). These learning models are executed and trained to predict software faults with higher accuracy. The study considers assessing the outcomes based on error-based performance metrics in detail to determine how well each learning model performs and how accurate it is at learning. It also looked at the factors contributing to the training loss of neural networks. The validation result demonstrates that, compared to logistic regression and support vector regression, neural network achieves a significantly lower error score for software fault prediction.</span>

Subtle defect detection on the surface of railway PCCS based on deep learning

The pantograph is a critical component of railway vehicles, and its operating status is directly related to railway operation safety. This paper concerns the detection of spall and crack on the surface of PCCS (Pantograph Carbon Contact Slide). For the detection of spall, EBF-YOLOv5 (YOLOv5 with Efficient Channel Attention, Bidirectional Feature Pyramid Network and Focal Loss) is put forward, which enhances the detection accuracy, reduces the omission rate and false detection rate of subtle spall on the surface of PCCS. For cracks segmentation, detect the crack at patch-level firstly, and then CBAMU-Net (Convolutional Block Attention Module U-Net) is presented to realize crack segmentation at pixel-level. The experimental results show it can effectively preserve the crack details to avoid over-segmentation and under-segmentation. In addition, the length and width of the crack are estimated on the segmented mask images, and the relative error is less than 7.1%. Experiments demonstrate that EBF-YOLOv5 and CBAMU-Net can meet the demand of practical application of typical PCCS subtle surface defects.

Reverse Engineering of Generative Models: Inferring Model Hyperparameters from Generated Images.

State-of-the-art (SOTA) Generative Models (GMs) can synthesize photo-realistic images that are hard for humans to distinguish from genuine photos. Identifying and understanding manipulated media are crucial to mitigate the social concerns on the potential misuse of GMs. We propose to perform reverse engineering of GMs to infer model hyperparameters from the images generated by these models. We define a novel problem, "model parsing", as estimating GM network architectures and training loss functions by examining their generated images - a task seemingly impossible for human beings. To tackle this problem, we propose a framework with two components: a Fingerprint Estimation Network (FEN), which estimates a GM fingerprint from a generated image by training with four constraints to encourage the fingerprint to have desired properties, and a Parsing Network (PN), which predicts network architecture and loss functions from the estimated fingerprints. To evaluate our approach, we collect a fake image dataset with 100K images generated by 116 different GMs. Extensive experiments show encouraging results in parsing the hyperparameters of the unseen models. Finally, our fingerprint estimation can be leveraged for deepfake detection and image attribution, as we show by reporting SOTA results on both the deepfake detection (Celeb-DF) and image attribution benchmarks.

Multi-Objective Reactive Power Optimization for Distribution Networks Based on the Modified Golden Eagle Algorithm

The Golden Eagle algorithm was proposed in 2020 by Abdolkarim et al. and has been applied in several fields. In this paper, we introduce the proposed algorithm to the reactive power optimization problem for distribution networks and develop a dual-objective reactive power optimization model with active network loss and node voltage bias. Logistic chaotic mapping and Levy flights are introduced to improve the original algorithm, which is prone to local optima and slow convergence. Finally, the IEEE30 node system is used for testing, which verifies the feasibility of the proposed algorithm in the reactive power optimization problem, and the improved algorithm has faster convergence and higher solution accuracy.

Multi-Task Probabilistic Regression With Overlap Maximization for Visual Tracking

Recent researches made a breakthrough in visual tracking accuracy. Many trackers benefit from the object state representations and network loss functions, which mine the output space and improve the power of supervision, respectively. Probabilistic regression method models the noises and uncertainties in the annotations. However, advanced trackers with probabilistic regression are not studied sufficiently in the aspect of supervision and the aspect of robustness of evaluation maximization. In this paper, an overlap maximization network in the manner of probabilistic regression is proposed to improve the learning ability of the network and the discriminative ability in the evaluation maximization. Firstly, the probabilistic regression is extended with the intersection over union (IoU) evaluation, which is normalized as a probability density in the regression space. Secondly, the classification probability is added as a branch of the iterative evaluation module to improve the ability of distinguishing objects in the evaluation maximization. Moreover, the two branches are constructed into a joint probabilistic regression task of IoU evaluation, which makes the network learn from two types of ground truth and provide a consistent result with multi-branch outputs. For feature interpretation, the strip pooling network and the space-time memory network are introduced to encode long-range context and provide dynamic features, respectively. Compared to the state-of-the-art probabilistic regression trackers and other advanced trackers, the experiments show that the proposed tracker achieves outstanding performance across the six datasets, including GOT-10k, LaSOT, TrackingNet, UAV123, OTB-100 and VOT2018.

Optimal Scheduling of New Energy and Thermal Power Combined Operation System

Nowadays, the power grid with a high proportion of renewable energy is increasingly popular. Taking a new energy power system including wind power (WP), photovoltaic power (PP), and pumped storage power (PSP) as an example, in order to solve the problem of resource waste such as massive WP-PP abandonment and power losses caused by the randomness and volatility of new energy, this paper establishes a combined power generation model of WP-PP-PSP-thermal power (TP). It achieves the goal of minimum network loss and minimum WP-PP abandonment in the combined power generation system according to the characteristics of PSP and considering various constraints and distribution network reconfiguration. Then, the second-order cone relaxation (SOCR) technology is used to transform the model into a mixed second-order cone programming model. Finally, the design scheme is tested by the improved IEEE 33-bus system, and the effectiveness of the algorithm is analyzed. The results show that peak cutting and valley filling can be realized when PSP energy is added to the WP-PP grid connection system.

Optimal Location of Electric Vehicle Charging Station in Reconfigured Radial Distribution Network

Electric vehicles are becoming increasingly popular because they are cleaner-burning and more efficient than combustion-engine automobiles. Due to the diminishing availability of fossil fuels and the carbon emissions produced by cars, electric vehicles (EV) have become a need for mobility in the near future. Electric car charging stations were established as a consequence of the increase in EVs. Electric car charging lowers voltage and increases real power loss in the radial distribution network. In order to mitigate real power loss and provide a stable voltage profile, the charging station has to be placed as efficiently as possible. The current research examines and elucidates the influence of electric vehicle charging stations (EVCS) on the balanced radial distribution network (BRDN) using particle swarm optimization (PSO). There are two steps involved in the planning process. The optimal placement for EVCS [1] in BRDN is identified in the first stage using PSO. The second stage uses reconfiguration to ensure higher EVCS stability. This two-step technique's main goal is to decrease the real power loss due to the placement of charging stations in the radial distribution network (RDN). The correctness of the suggested approach is expounded upon in four different cases. The second, third, and fourth cases use a two-step procedure, which is then compared to the base case. In the base case EVCS is ideally located in BRDN using PSO. The fourth case employs a concurrent two-step optimization approach and demonstrates superior performance compared to other cases. The suggested approach is implemented and evaluated on the IEEE 69 bus BRDN using the MATLAB software.

A granular framework for modeling the capacity loss and recovery of regional transportation networks under seismic hazards: A case study on the Port of Los Angeles

Earthquakes, being both unpredictable and potentially destructive, pose great risks to critical infrastructure systems like transportation. It becomes crucial, therefore, to have both a fine-grained and holistic understanding of how the current state of a transportation system would fare during hypothetical hazard scenarios. This paper introduces a synthesis approach to assessing the impacts of earthquakes by coupling an image-based structure-and-site-specific bridge fragility generation methodology with regional-scale traffic simulations and economic loss prediction models. The proposed approach’s use of context-rich data such as OpenStreetMap and Google Street View enables incorporating information that is abstracted in standard loss analysis tools like HAZUS in order to construct nonlinear bridge models and corresponding fragility functions. The framework uses a semi-dynamic traffic assignment model run on a regional traffic network that includes all freeways and local roads (1,444,790 edges) and outputs traffic volume on roads before and after bridge closures due to an earthquake as well as impacts to individual trips (42,056,426 trips). The combination of these models enables granularity, facilitating a bottom-up approach to estimating costs incurred solely due to damages to the transportation network. As a case study, the proposed framework is applied to the road network surrounding the Port of Los Angeles—an infrastructure of crucial importance—for assessing resilience and losses at a high resolution. It is found that the port area is disproportionately impacted in the hypothetical earthquake scenario, and delays in bridge repair can lead to a 50% increase in costs.

A resilient self‐healing approach for active distribution networks considering dynamic microgrid formation

AbstractA large number of renewable distributed generation (DG) systems connect to the distribution network, affecting the structure of the traditional distribution network and forming an active distribution network (ADN). Although the accelerating grid penetration of DGs brings significant challenges to the distribution network operation, the islanded operation capability of DGs provides a flexible solution to the self‐healing of ADNs from faults. To ensure that ADN can quickly recover and reconfigure in the event of a fault and continue to maintain safe, economical, and reliable operation, this paper proposes a dynamic microgrid formation method for ADNs combined with the dynamic network reconfiguration and intentional islanding operation of DGs. An optimization model is designed to represent the proposed self‐healing method, maximizing the load restoration and minimizing the DG cost, line network loss, and voltage excursion simultaneously. A binary hybrid optimization solver is applied to pursue the optimal self‐healing schedules from the optimization model. The self‐healing method is evaluated on the Institute of Electrical and Electronics Engineers (IEEE) 33‐node system and the IEEE 123‐node system, which indicate its rationality and effectiveness fully verified. By optimizing the on and off states of the normally open switches and the on‐grid and off‐grid operation states of DGs, ADNs not only get healed with minimum load curtailment, but also achieve minimal DG generation cost, network loss, and node voltage deviation. In addition, compared with traditional solvers, the proposed solver has a slightly higher computational time than the corresponding solver.

MSA R-CNN: A comprehensive approach to remote sensing object detection and scene understanding

Remote sensing scene understanding is crucial for extracting valuable information from high-resolution images, including object detection and classification. Traditional object detection methods face challenges in handling the diverse scales, orientations, and complex backgrounds present in remote sensing data. In this paper, we propose a novel remote sensing scene understanding system called multiscale-attention R-CNN (MSA R-CNN), which incorporates a super multiscale feature extraction network (SMENet) for enhanced feature extraction from multiscale images, an adaptive dynamic inner lateral (ADIL) connection module to tackle information loss in feature pyramid networks (FPN), and a distributed lightweight attention module (DLAM) to refine feature information processing. Furthermore, a new dataset combining the DIOR and DOTA datasets is introduced to extract the background information of detected objects and evaluate the proposed system’s performance. MSA R-CNN achieved an mAP of 74.37% on the DIOR dataset when the gamma value was set to 0.2 and 81.97% on the DOTA dataset when the gamma value was set to 0.1 with the same learning rate, outperforming state-of-the-art models on both datasets. The proposed system demonstrates significant improvements in both object detection and background information extraction, providing a comprehensive solution for remote sensing scene understanding.

Optimizing Multimedia Data Delivery in Wireless Multimedia Sensor Network: A QoS-Driven Routing Protocol Design

In recent years, there has been a shift towards utilizing wireless sensor networks (WSN) for multimedia applications, even though these networks were originally designed for transmitting small data volumes. In order to satisfy the increasing need for improved throughput and real-time performance when handling multimedia content, it becomes essential to streamline and enhance the routing protocols. In this document, we introduce a routing algorithm called Honey Badger optimization Algorithm (HBOA). This algorithm is designed to minimize end-to-end packet delay and reduce packet loss in wireless sensor networks (WSN). Instead of selecting shortest path to the destination HBOA selects optimised high throughput path to transport multimedia packets ovet the nertwork. Furthermore, it incorporates a network void-bypass mechanism to enhance network reliability when it encounters gaps in the network. Energy management is integrated to enhance the longevity of sensor nodes by adjusting the radio transmission coverage to a distance that sensor nodes can efficiently reach. The simulation results when compared with existing routing protocols demonstrate that HBOA achieves a 8% reduction in end-to-end delay and a 18% decrease in packet loss ratio while effectively managing energy consumption when compared to alternative algorithms. Load balancing occurs as a result of energy management, which makes the remaining energy of sensor nodes uniformly dispersed throughout the entire network. Through Simulation, the performance of HBOA is also compared to that of AGEM, TPGF, GPSR, and AODV.the esults suggests the HBOA performance better under conditions that are similat to these, such as low packet end-to-end- delay, low packet loss, and load balancing of routing channels.

Securing smart home against sinkhole attack using weight‐based IDS placement strategy

AbstractExtensive use of the Internet of Things (IoT) in smart homes makes users' lives easy and comfortable. Yet, these resource‐constrained devices are prone to manifold security attacks. The sinkhole attack is one of the most destructive attacks that disrupt smart home operations, causing user dissatisfaction. Existing intrusion detection systems (IDS) cannot handle sinkhole attacks competently as they (i) do not consider the node capacity for being an IDS agent, leading to a low attack detection ratio, (ii) do not examine the sinkhole node's role when mitigating attacks, causing remaining network disconnection with the root node and (iii) do not consider replacing energy‐exhausted IDS nodes, causing connectivity loss of partial network with the root. This paper addresses these shortcomings and adequately presents a mechanism to handle sinkhole attacks. A formulation for assigning weights to network nodes based on their resources is proposed here. An IDS placement strategy is introduced to place IDS agents on particular resourceful nodes that extend network lifetime and enhance attack detection capability. We present a novel attack detection and mitigation strategy by ensuring network connectivity. The proposed mechanism achieves 95% attack detection accuracy and reduces false negative rates by 25% and energy consumption reasonably compared to the state‐of‐the‐art.

Distributed Power High Dynamic Group Control Technology and Grid Connection Demonstration

Abstract Currently, high penetration distributed power generation group control technology makes it difficult to effectively regulate complex power grids. Therefore, on the ground of the division of distributed photovoltaic power generation clusters, this study proposes a dual-layer control model for distribution networks that integrates the cuckoo particle swarm algorithm and verifies its effectiveness. The experiment showcases that the standard deviations of the research algorithm on the four benchmark functions under the 50 dimension are 2.95e-02, 6.73e+02, 6.48e-04, and 6.45e-06, respectively, most of which are lower than the comparison algorithm. On the basis of connecting to the photovoltaic system, the voltage amplitude fluctuation of the double-layer control model designed through research using control algorithms is relatively small, basically maintaining around 1.05. In addition, after implementing control in practical applications, the actual line loss of the distribution network decreased from the maximum of 15.3% to 8.9%, meeting the line loss requirements of the actual distribution network, indicating the effectiveness of the double-layer control model. Meanwhile, after applying the double-layer control model, the voltage fluctuation of node 42 significantly decreased, and the numerical value remained at 1.05 p.u., achieving a good voltage condition effect. The voltage fluctuation of grid connection in distributed power generation has been significantly improved when applied to the actual grid connection. Overall, the proposed dual-layer control model for the distribution network is effective and practical, which can effectively promote the construction of practical grid connection demonstration projects.

A failure or success?: A qualitative analysis of post-disaster relocation among survivors of 2010 floods in Pakistan

In Pakistan, the enormous 2010 floods inflicted unprecedented destruction to 0.5 million houses in South Punjab. This study was conducted in Muzaffargarh, a district of the Punjab province of Pakistan, one of the worst affected areas during the 2010 floods. As an adaptation and disaster risk reduction measure, an International NGO in collaboration with the Government of the Punjab province constructed 1274 houses for flood-affected communities. The houses were allocated through a lottery method. Initially, the allottees moved to the new settlements for the sake of possession. However, over time, it was observed that many flood-affected individuals sold out or rented their houses and returned to their villages/settlements. This qualitative study was conducted to identify the factors affecting the flood survivors’ decision to return to their original villages/settlements after a certain time. The specific objective of the study was to identify the factors that pushed and pulled the flood survivors into their original settlements. The current study interviewed 20 respondents through the snowball sampling technique by using an interview guide. The findings highlighted the loss of social capital, place attachment, unavailability of grazing places, non-transfer of house, and increase in the cost of living were the major factors that pushed the flood survivors to return to their native villages. The current study highlighted social and psychological costs (such as nostalgia, homesickness, loss of social networks, etc.) for the relocation of disaster survivors. This study contains important policy implications regarding relocation decisions taken by the government and policymakers. Findings indicate relocation decisions as the failure to address the local needs, priorities, and contexts.

VGAN-BL: imbalanced data classification based on generative adversarial network and biased loss

VGAN-BL: imbalanced data classification based on generative adversarial network and biased loss

Active distribution network operational optimization problem: A multi-objective tuna swarm optimization model

This study proposes multi-objective tuna swarm optimization through multi-objective transformation, initialization improvement and population variation for active distribution network (ADN). The ADN energy optimization model with dynamic reconfiguration, reactive power compensation, on-load tap changer and controllable load coordinated control are established with the minimum economic and environmental cost. Several controllable resources increase and increase the complexity of energy optimization problem in realized the effective clean energy consumption and the power grid stable operation. The minimum control cost and the minimum node voltage deviation are proposed. This study also proposes a decision-making method based on pareto front and an index to evaluate the maximum extensible dimension of intelligent algorithms in the process of analyzing energy optimization problems with intelligent algorithms. The proposed ADN energy optimization method based on multi-objective tuna swarm optimization shows excellent results after testing with the improved IEEE33 system. The voltage deviation and network loss are reduced by 51.62% and 22.16% on average compared with the single means control. The proposed model provides the support for the new energy consumption and the power grid stable operation, has strong engineering significance in the intelligent upgrading and power grid active control transformation.

Comprehensive energy efficiency optimization algorithm for steel load considering network reconstruction and demand response

Industrial loads are usually energy intensive and inefficient. The optimization of energy efficiency management in steel plants is still in the early stage of development. Considering the topology of power grid, it is an urgent problem to improve the operation economy and load side energy efficiency of steel plants. In this paper, a two-level collaborative optimization method is proposed, which takes into account the dynamic reconstruction cost, transmission loss cost, energy cost and demand response benefit. The upper level objective is the optimization of topology in the grid structure to optimize the power loss and dynamic reconstruction costs of the grid. The lower level is the energy cost considering demand response, real time price and dynamic demand response price. Firstly, the mathematical models of stable load, impact load and the steel production line load are built. The key parameters are identified by the Back Propagation neural network algorithm according to the actual production data. Secondly, considering the constraints of grid structure and load operation capacity, the impact of dynamic grid loss and real-time dynamic electricity price on the energy efficiency of the whole grid are analyzed in depth. The optimal operation model considering the dynamic reconfiguration and grid tramission loss of distribution network is built. Taking a steel plant park in Northeast China as an example, it is proved that the optimization model can improve energy efficiency on the load side by optimizing energy consumption and demand response participation time on load side. The energy cost is reduced by 17.77% on the load side, the network loss is reduced by 1.8%, and the operating cost of the power grid is reduced by 26.2%, which has a positive effect on improving energy utilization efficiency, reducing distribution network loss, and improving overall economic efficiency.

A new continuous variables and convex relaxation based formulation for reconfiguration of unbalanced mutually coupled active distribution networks

With the integration of renewable energy-based sources and demand response initiatives, the distribution network has transformed into an active network with bidirectional flows. Network Reconfiguration (NR), i.e., changing network topology, is an alternative to minimize power loss in the distribution network with sectionalizing switches (normally closed) and tie-line switches (normally open). Due to the discrete nature of switches (on or off), the NR formulation is generally a mixed-integer NP-hard nonlinear problem. This paper proposes the first NR formulation for an unbalanced active distribution network with continuous variables to indicate switches’ statuses in NR while considering mutual coupling between line impedances. The obtained solution provides the optimal real power and voltage magnitude settings of dispatchable Distributed Generators (DGs) and network topology for the minimum active power loss in the network. The proposed formulation considers all practical aspects of the distribution network, i.e., mutual coupling, load unbalances, angular unbalance, and the absence of one or two phases in a line section, which, otherwise, have not been considered together in previous formulations. Further, a novel convex relaxation for bus voltage phase angles avoids explicit voltage recovery. Compared to prior formulations, numeric results for unbalanced radial distribution networks showcase the proposed formulation’s effectiveness.

HairManip: High quality hair manipulation via hair element disentangling

Hair editing is challenging due to the complexity and variety of hair materials and shapes. Existing methods employ reference images or user-painted masks to edit hair and have achieved promising results. However, discrepancies in color and shape between the source and target hair can occasionally result in unrealistic results. Therefore, we propose a new hair editing method named HairManip, which decouples the hair information from the input source image into shape and color components. We then train hairstyle and hair color editing sub-networks to handle this complex information independently. To further enhance editing efficiency and accuracy, we introduce a latent code preprocessing module that effectively extracts meaningful features from hair regions, thereby improving the model’s editing capabilities. The experimental results demonstrate that our method achieves significant results in editing accuracy and authenticity, thanks to the carefully designed network structure and loss functions. Code can be found at https://github.com/Zlin0530/HairManip.

Strategic Decision-Making in Sustainable Water Management Using Demand Analysis and the Water Evaluation and Planning Model

Water infrastructure management relies on information, communication strategies, and affordable technologies. This paper used demand analysis and modeling to guide strategic decision-making in sustainable water management for the urban cluster in Tlemcen, Algeria. To achieve this, the water supply and demand of the study area were assessed over the past three decades. The Water Evaluation and Planning (WEAP) system was employed considering different future scenarios to help decision-makers consider the best choices for sustainable urban water resources management. The results showed that the average water production and distribution efficiency was only 46% due to the high network loss. Therefore, urgent action should be considered to increase the efficiency of the distribution network. Moreover, the outcome showed severe unmet demand in 2050, which can be managed by improving the water networks, increasing conventional water production, and reducing personal water consumption. In cooperation with key stakeholders, new scenarios can be analyzed to develop efficient water management policies and to implement sustainable water allocation approaches.